---
license: apache-2.0
base_model: mistralai/Mistral-Nemo-Base-2407
tags:
- general-purpose
- text-generation
---

# Astra-v1-12B

Astra-v1-12B is a fine-tuned version of the base model [Mistral-Nemo-Base-2407](https://huggingface.co/mistralai/Mistral-Nemo-Base-2407), developed for general-purpose natural language processing tasks. It was fine-tuned to replicate the quality and style of Claude 3's Sonnet and Opus models.

![Astra-v1-12B](https://i.imgur.com/rCXcyno.png)


### Model Description

Astra-v1-12B is a general-purpose transformer-based language model fine-tuned for instruction-following tasks. The fine-tuning was designed to match the high-quality generation seen in Claude 3's Sonnet and Opus models, optimized for tasks such as text generation, summarization, question answering, and more.

- **Developed by:** P0x0
- **Finetuned from:** [Mistral-Nemo-Base-2407](https://huggingface.co/mistralai/Mistral-Nemo-Base-2407)
- **License:** Apache 2.0

### Model Sources

- **Repository:** [https://huggingface.co/P0x0/astra-v1-12b](https://huggingface.co/P0x0/astra-v1-12b)
 - **GGUF** [Astra-v1-12B-GGUF](https://huggingface.co/P0x0/Astra-v1-12B-GGUF)

## Uses

### Direct Use

Astra-v1-12B can be used directly for a wide range of NLP tasks, including:
- Text generation
- Summarization
- Question answering
- Dialogue systems

### Out-of-Scope Use

Astra-v1-12B is not intended for real-time decision-making in critical applications or generating harmful or biased content.
## How to Get Started with the Model

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("P0x0/astra-v1-12b")
model = AutoModelForCausalLM.from_pretrained("P0x0/astra-v1-12b")

input_text = "Explain the theory of relativity in simple terms."
inputs = tokenizer(input_text, return_tensors="pt")
outputs = model.generate(**inputs)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```
I encourage you to provide feedback on the model's performance. If you'd like to create your own quantizations, feel free to do so and let me know how it works for you!

model-index:
  - name: P0x0/Astra-v1-12B
    results:
      - task:
          type: text-generation
        metrics:
          - name: Average
            type: Average
            value: 19.46
          - name: IFEval
            type: IFEval
            value: 28.06
          - name: BBH
            type: BBH
            value: 31.81
          - name: MATH Lvl 5
            type: MATH Lvl 5
            value: 9.67
          - name: GPQA
            type: GPQA
            value: 8.5
          - name: MUSR
            type: MUSR
            value: 11.38
          - name: MMLU-PRO
            type: MMLU-PRO
            value: 27.34